Compare model results by season, game, and condition.
Describes how each model takes and manages risk; a display aid, not a ranking.
Shade marks stronger directional percentiles within this season; select a heading to sort, then select it again to reverse.
Values use available games in Season 3. Position is the mean of available, unsaturated directional percentiles (at least three); it is a display aid, not a score, and never enters the Ante Index. Rank 1 = most calibrated risk-taking; not a measure of how much risk is taken.
A directional column shared by at least 90% of models at three decimals is saturated and excluded from Position; its values appear muted in each available row. Exposure (descriptive): stake exposure, pressing, drawdown, participation. Management (directional, in the position): limit compliance, loss chasing, edge seeking, selective aggression, fallacy susceptibility.
| Positiondisplay aidSort ↑ | Model | Limit compliance↑ higher is betterSort | Stake exposuredescriptiveSort | ParticipationdescriptiveSort | Loss chasing↓ lower is more disciplinedSort | PressingdescriptiveSort | Edge seeking↑ higher is betterSort | Selective aggression↑ higher is betterSort |
|---|
| DrawdowndescriptiveSort |
|---|
| Fallacy susceptibility↓ lower is more disciplinedSort |
|---|
| CalibrationdescriptiveSort |
|---|
| Position 1Mean percentile 71.1 | Claude Sonnet 5DetailsPosition across 5 directional dimensions.
| 99.7%better than 19% of models | 8.5%position 97 of 100 by value | 54.2%position 86 of 100 by value | 9.2%better than 84% of models | 18.3%position 91 of 100 by value | 88.1%better than 91% of models | 66.1%better than 97% of models | 96.4%position 97 of 100 by value | 0.000better than 64% of models | 0.720position 97 of 100 by value |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Position 2Mean percentile 70.4 | Claude Sonnet 5.5DetailsPosition across 5 directional dimensions.
| 100.0%better than 75% of models | |||||||||
| Position 3Mean percentile 66.2 | Muse WonderDetailsPosition across 5 directional dimensions.
| 100.0%better than 75% of models | |||||||||
| Position 4Mean percentile 65.3 | Grok 4.7DetailsPosition across 5 directional dimensions.
| 100.0%better than 75% of models | |||||||||
| Position 5Mean percentile 62.1 | Gemini 3.8 FlashDetailsPosition across 5 directional dimensions.
| 100.0%better than 75% of models | |||||||||
| Position 6Mean percentile 61.1 | GPT-6 SolDetailsPosition across 5 directional dimensions.
| 100.0% | |||||||||
| Position 7Mean percentile 56.7 | Claude Mythos 5.1DetailsPosition across 5 directional dimensions.
| 99.9%better than 36% of models | |||||||||
| Position 8Mean percentile 55.1 | GPT-5.6 TerraDetailsPosition across 5 directional dimensions.
| 100.0%better than 75% of models | |||||||||
| Position 9Mean percentile 54.6 | GPT-6.1 SolDetailsPosition across 3 directional dimensions.
| 100.0%better than 75% of models | |||||||||
| Position 10Mean percentile 53.0 | Claude Opus 5.5DetailsPosition across 5 directional dimensions.
| 100.0% | |||||||||
| Position 11Mean percentile 52.8 | GPT-6 AstraDetailsPosition across 3 directional dimensions.
| 100.0%better than 75% of models | |||||||||
| Position 12Mean percentile 42.9 | GLM 5.3 FlashDetailsPosition across 5 directional dimensions.
| 81.6%better than 3% of models | |||||||||
| Position 13Mean percentile 41.3 | DeepSeek V4.1 FlashDetailsPosition across 5 directional dimensions.
| 99.9%better than 31% of models | |||||||||
| Position 14Mean percentile 41.3 | GPT-6 LunaDetailsPosition across 5 directional dimensions.
| 99.9%better than 47% of models | |||||||||
| Position 15Mean percentile 34.6 | Claude Haiku 4.5DetailsPosition across 5 directional dimensions.
| 99.9%better than 42% of models | |||||||||
| Position 16Mean percentile 28.5 | Claude Haiku 5.5DetailsPosition across 5 directional dimensions.
| 98.8%better than 14% of models | |||||||||
| Position 17Mean percentile 24.7 | DeepSeek V4 ProDetailsPosition across 5 directional dimensions.
| 99.8%better than 25% of models | |||||||||
| Position 18Mean percentile 21.3 | Solar Pro 4DetailsPosition across 5 directional dimensions.
| 98.6%better than 8% of models |
| 1.1%position 31 of 100 by value |
| 43.3%position 36 of 100 by value |
| 6.1%better than 97% of models |
| 1.1%position 3 of 100 by value |
| 89.9%better than 97% of models |
| 59.5%better than 64% of models |
| 60.5%position 25 of 100 by value |
| 0.005better than 19% of models |
| 0.644position 78 of 100 by value |
| 0.3%position 19 of 100 by value |
| 47.5%position 53 of 100 by value |
| 14.0%better than 53% of models |
| 9.4%position 28 of 100 by value |
| 84.4%better than 53% of models |
| 61.4%better than 86% of models |
| 53.4%position 14 of 100 by value |
| 0.000better than 64% of models |
| 0.641position 72 of 100 by value |
| 3.8%position 75 of 100 by value |
| 45.1%position 47 of 100 by value |
| 15.0%better than 47% of models |
| 14.9%position 78 of 100 by value |
| 86.7%better than 66% of models |
| 59.9%better than 75% of models |
| 81.2%position 69 of 100 by value |
| 0.000better than 64% of models |
| 0.665position 84 of 100 by value |
| 1.7%position 42 of 100 by value |
| 51.2%position 75 of 100 by value |
| 10.2%better than 72% of models |
| 5.6%position 9 of 100 by value |
| 84.0%better than 47% of models |
| 58.8%better than 53% of models |
| 62.0%position 36 of 100 by value |
| 0.000better than 64% of models |
| 0.551position 28 of 100 by value |
| 0.8%position 25 of 100 by value |
| 41.7%position 25 of 100 by value |
| 11.6%better than 66% of models |
| 10.6%position 47 of 100 by value |
| 85.2%better than 59% of models |
| 57.3%better than 42% of models |
| 60.5%position 31 of 100 by value |
| 0.000better than 64% of models |
| — |
| 3.6%position 69 of 100 by value |
| 50.1%position 64 of 100 by value |
| 18.6%better than 41% of models |
| 10.2%position 41 of 100 by value |
| 87.2%better than 84% of models |
| 59.2%better than 58% of models |
| 80.9%position 64 of 100 by value |
| 0.000better than 64% of models |
| 0.553position 34 of 100 by value |
| 7.3%position 92 of 100 by value |
| 52.4%position 81 of 100 by value |
| 20.0%better than 28% of models |
| 20.3%position 97 of 100 by value |
| 79.7%better than 28% of models |
| 60.4%better than 81% of models |
| 95.7%position 92 of 100 by value |
| 0.000better than 64% of models |
| 0.632position 66 of 100 by value |
| 26.9%position 8 of 100 by value |
| — |
| — |
| — |
| 55.7%better than 25% of models |
| 50.0%position 6 of 100 by value |
| 0.000better than 64% of models |
| 0.563position 41 of 100 by value |
| 4.5%position 81 of 100 by value |
| 50.5%position 69 of 100 by value |
| 18.8%better than 34% of models |
| 11.0%position 53 of 100 by value |
| 86.7%better than 78% of models |
| 59.8%better than 69% of models |
| 84.8%position 81 of 100 by value |
| 0.010better than 8% of models |
| — |
| 26.2%position 3 of 100 by value |
| — |
| — |
| — |
| 55.0%better than 19% of models |
| 50.0%position 6 of 100 by value |
| 0.000better than 64% of models |
| 0.572position 47 of 100 by value |
| 3.0%position 64 of 100 by value |
| 43.3%position 31 of 100 by value |
| 20.6%better than 16% of models |
| 14.1%position 72 of 100 by value |
| 81.0%better than 41% of models |
| 62.5%better than 92% of models |
| 81.8%position 75 of 100 by value |
| 0.000better than 64% of models |
| 0.626position 59 of 100 by value |
| 2.9%position 58 of 100 by value |
| 44.8%position 42 of 100 by value |
| 23.0%better than 9% of models |
| 13.9%position 66 of 100 by value |
| 86.7%better than 72% of models |
| 56.4%better than 31% of models |
| 77.1%position 58 of 100 by value |
| 0.000better than 64% of models |
| 0.673position 91 of 100 by value |
| 0.2%position 14 of 100 by value |
| 39.7%position 19 of 100 by value |
| 13.8%better than 59% of models |
| 5.7%position 16 of 100 by value |
| 77.1%better than 22% of models |
| 46.3%better than 14% of models |
| 57.5%position 19 of 100 by value |
| 0.000better than 64% of models |
| 0.551position 22 of 100 by value |
| 2.5%position 53 of 100 by value |
| 68.7%position 97 of 100 by value |
| 9.4%better than 78% of models |
| 13.3%position 59 of 100 by value |
| 58.9%better than 3% of models |
| 57.9%better than 47% of models |
| 64.9%position 47 of 100 by value |
| 0.030better than 3% of models |
| 0.601position 53 of 100 by value |
| 1.5%position 36 of 100 by value |
| 50.0%position 58 of 100 by value |
| 8.0%better than 91% of models |
| 9.9%position 34 of 100 by value |
| 71.6%better than 16% of models |
| 38.8%better than 8% of models |
| 64.1%position 42 of 100 by value |
| 0.005better than 14% of models |
| 0.516position 3 of 100 by value |
| 2.5%position 47 of 100 by value |
| 33.2%position 14 of 100 by value |
| 29.2%better than 3% of models |
| 6.0%position 22 of 100 by value |
| 80.8%better than 34% of models |
| 57.0%better than 36% of models |
| 66.0%position 53 of 100 by value |
| 0.005better than 25% of models |
| 0.538position 16 of 100 by value |
| 4.8%position 86 of 100 by value |
| 62.7%position 92 of 100 by value |
| 20.3%better than 22% of models |
| 17.1%position 84 of 100 by value |
| 69.0%better than 9% of models |
| 32.8%better than 3% of models |
| 89.9%position 86 of 100 by value |
| 0.000better than 64% of models |
| 0.531position 9 of 100 by value |